A single oncology patient's record can run into the thousands of pages - pathology reports, clinical notes, genomic panels, and scanned faxes. For years, cancer centers have paid highly trained staff to read all of it by hand to answer one question at a time: Does this patient qualify for a trial? What's their current disease status? Which registry field goes where? Triomics is trying to make that manual reading obsolete.
Founded in 2021 by Sarim Khan and Hrituraj Singh, Triomics builds task-driven AI agents that embed directly into clinical workflows and process medical records at scale, in real time. The company describes itself plainly - "the AI infrastructure for modern oncology" - and the framing is deliberate. This is not a chatbot bolted onto an EHR. It is a layer meant to sit underneath the work oncology teams already do.
The two founders were college friends before they became, respectively, an MIT biotech researcher and an Adobe AI researcher. That combination - clinical data fluency plus applied machine learning - shows up in the product's core conviction: general-purpose models write summaries, but oncology needs something that can reason across an entire chart and cite exactly where each answer came from.
Cancer care is, by the founders' own account, the hardest place to build AI. Records are long, unstructured, and inconsistent. The stakes are clinical. And clinicians will not trust a black box. Triomics' answer is an oncology-specific model, OncoLLM, that returns auditable, cited responses - every claim linked back to the pathology, molecular data, or note it was drawn from.
By 2026 the approach had reached 4 of the top 10 U.S. News Best Hospitals for Cancer, including Memorial Sloan Kettering, MD Anderson, Yale Cancer Center, and Mount Sinai - alongside several of the largest community oncology practices. In May 2026, Battery Ventures led a $22 million Series B to scale it further.
